EP0379491B1 - Pitch-control system - Google Patents

Abstract

A pitch-control system automatically corrects the pitch in accordance with a harmony-dependent, variable key, in particular the harmonic key, for a musical instrument having an input device for inputting note input signals in a predetermined fixed key, in particular the tempered key, and a sound-generating device to which the note input signals are applied. The pitch-control system comprises a chord-recognition circuit which determines, for each input signal pattern corresponding to a chord, whether the input signal pattern corresponds to a chord pattern in a predeterined plurality of chord patterns, a signal pattern memory circuit (34) which records a signal pattern for each chord pattern in the predetermined plurality of chord patterns, and a control circuit (32) which, when the chord recognition circuit determines that an input singal pattern corresponding to one of the predetermined chord patterns is applied, causes the signal pattern memory circuit (34) to apply the signal pattern corresponding to the chord pattern so determined to the tone-generating device (20) for generation of the chord corresponding to the variable key.

Description

• a) bei einem einem Akkord entsprechenden Eingabesignalmuster stellt man durch Vergleich mit Musterakkorden einer vorgegebenen Menge an Akkordmustern fest, ob eines dieser Akkordmuster vorliegt;
• b) bei Vorliegen eines Akkordmusters wird das Eingabesignalmuster durch ein entsprechend diesem Akkordmuster korrigiertes Eingabesignalmuster ersetzt und an die Ton-Erzeugungseinrichtung angelegt, wobei man das einem vorbestimmten Grundton des jeweiligen Akkordmusters zugeordnete Signal der Signalmuster entsprechend der vorgegebenen festen Stimmung festlegt und die den übrigen Tönen des Akkordmusters zugeordneten Signale des Signalmusters, vom Grundton ausgehend, entsprechend der variablen Stimmung korrigiert.

The invention relates to a method for automatic pitch correction according to a harmony-dependent variable mood, in particular the harmonic mood, for a musical instrument with an input device for inputting note input signals in a predetermined fixed mood, in particular the equally tempered mood, and with a tone generating device to the the note input signals can be applied with the following steps:

• a) in the case of an input signal pattern corresponding to a chord, it is determined by comparison with pattern chords of a predetermined amount of chord patterns whether one of these chord patterns is present;
• b) in the presence of a chord pattern, the input signal pattern is replaced by an input signal pattern corrected in accordance with this chord pattern and is applied to the tone generating device, the signal of the signal pattern associated with a predetermined fundamental of the respective chord pattern being determined in accordance with the predetermined fixed mood and the other tones of the Chord pattern-assigned signals of the signal pattern, starting from the fundamental, corrected according to the variable tuning.

The long-known problem with the choice of tuning is that the "harmonically pure tuning" preferred in the course of the development of polyphony produces particularly pleasant chord sounds due to the partial agreement of harmonics and fundamental vibrations of the chord tones; however, the transition from one key to another requires a corresponding adjustment of the tuning (even within a harmonically tuned key there are chords with a frequency ratio that does not correspond to the harmonically pure tuning). In order to enable instruments that cannot change their mood while playing (e.g. the keyboard instruments piano or organ), to play in different keys and to modulate from one key to another, these instruments are tuned in a predetermined fixed mood, in which the chords sound more or less equally good (or equally bad) in the key in question. An example of such a fixed mood is the tempered mood according to Johann Sebastian Bach. However, other fixed moods have been proposed, in particular the baroque moods "Werckmeister" and "Kirnberger" (see DE-C 25 58 716), which prefer certain chords or keys, but at the expense of other chords or keys.

From DE-A-33 04 995 it is known to provide an electronic keyboard instrument with tonality selection keys to be operated manually during play, the actuation of which has the result that the keyboard instrument is currently harmoniously pure with respect to the selected tonality (eg C major Subdominant) is tuned. This manual operation disrupts the flow of the game and also assumes that the player immediately recognizes the respective tonality during the game.

From DE-A-35 45 986 an electronically controlled musical instrument is known which checks successive tones to determine whether they can be assigned to a key. If this is the case, the instrument is then harmoniously tuned to the corresponding key, eg C major or E minor. The seven are used to clearly identify the respective key Tones of an octave needed. The key identification may therefore only take place after a relatively long game or not at all. Accordingly, if a piece of music starts again, or if modulation or backing takes place, the chords struck will not sound harmoniously. Even transitional notes away from the key, such as those that occur with decorations or chromatic passages, cause such an instrument to temporarily initiate the tonally determined harmonic attunement and to harmonize with the old or a new tonality only after a waiting period, if necessary. Likewise, such an instrument does not tune to a key or continuously changes when polyphonic music sounds that cannot be fixed to a major or minor key.

From DE-B-30 23 578 a circuit arrangement for identifying the chord type and its root note is known, which serves to generate an automatic accompaniment to a melody voice played on the instrument. WO-A-80/00110 also shows a circuit arrangement for chord recognition with the possibility of sounding the entire chord by pressing the key assigned to a chord root.

From US-A-4 248 119 a method of the type mentioned is known, in which after identification of the respective chord, the chord tones are corrected in relation to the chord root so that harmonically pure chord tuning results. The chord root note is not corrected, ie it corresponds to the specified fixed (equally tempered) mood. With this known method, harmonically pure chords result. However, it was recognized that different chords played in immediate succession may sound uncomfortable especially when the same tones occur in both chords, but in different functions. Due to the chord-specific and therefore different frequency correction of these tones, they lie at different pitches in both chords, which is perceived as unpleasant with a correspondingly large frequency difference.

In contrast, the object of the invention is to disclose a method of the type mentioned at the outset with which a further improvement in sound can be achieved.

This object is achieved by additionally correcting the signal of the input signal pattern assigned to the fundamental tone with respect to the predetermined fixed tuning, and correcting the signals of the input signal pattern assigned to the other tones of the chord pattern, starting from the corrected fundamental tone, in accordance with the variable tuning, whereby this is the fundamental tone a signal associated with a chord is corrected such that the correspondingly corrected tone emitted by the tone generating device is higher or lower than the basic tone in the predetermined fixed tuning, depending on whether the chord tones corrected in accordance with the input signal pattern are on average lower or higher than the uncorrected Chord tones in the solid mood.

This measure according to the invention results in a kind of alignment of successive chords with the result that the frequency differences of the same tones are reduced. Successive chords sound noticeably better.

The signal assigned to the fundamental tone of a chord is particularly preferably corrected in such a way that the shift in an average frequency of the chord tones due to the correction of the chord tones is at least approximately compensated for by the signal pattern. In the case of using correction signals indicating relative frequency changes It is proposed that the signal associated with the fundamental tone of a chord be corrected with an additional correction signal indicating a relative frequency change, which corresponds to the mean value of the correction signals for the input signals which also indicate relative frequency change, but with the opposite sign.

With this additional correction, notes of the same name in chords that can be represented as steps in a single key are only retuned to such an extent that this retuning is below the audible limit. An example would be the tone E in D major, as a third of level I, C-E-G, as a root of level III, E-G-H or as a fifth of level VI, A-C-E. So there is a key-independent but key-friendly change of heart.

In the case of small major seventh chords, special attention must be paid to the tone with the function of the minor seventh in the chord. In a tempered mood, this note has a value of 1000 cents in relation to the root note of the chord. In historical times, the minor seventh of a scale was often tuned to the value 7/4 to the fundamental, which corresponds to the frequency ratio of the 7th overtone of a natural tone series. However, this value is unusable for today's music practice, since at 969 cents it is too far from the value of the tempered mood.

The frequency ratio offers a useful value, as occurs in the dominant seventh chord within a harmonically tuned major scale. In the dominant seventh chord, the root note is formed by the fifth of the key concerned, while the minor seventh is formed by the fourth of the octave above. The fifth has a frequency ratio of 3/2 to the key of the key, the fourth above it has a ratio of the next octave from 2x4 / 3 = 8/3 to the key of the key. So the minor seventh of the dominant seventh chord is 8/3: 3/2 = 16/9. This corresponds to 997 cents to the chord root. After the additional correction already mentioned, this tone in the small major seventh chord has a correction of preferably +1 cent to the frequency of the tempered mood, which sounds good.

With a small minor seventh chord, on the other hand, the tone with the function of the minor seventh in the chord has the usual value of 6: 5 to the fifth, which corresponds to a persistence of 1018 cents to the root.

In order to only have to make a chord pattern comparison in the 12-tone space corresponding to an octave, it is proposed that the input signal pattern be projected onto a predetermined octave (definition octave) and compared with the chord patterns of the predetermined amount of chord patterns, which are also limited to one octave .

In a first alternative, this comparison can be made by shifting the input signal pattern within the definition octave as a whole in semitones and counting the shift steps until a signal is at a predetermined end of the definition octave, and by including the signal pattern shifted in this way compares the chord patterns to the predetermined amount of chord patterns, with each chord pattern also having a chord note at the predetermined end of the octave.

However, one can also proceed in such a way that the chord patterns of the specified number of chord patterns within the octave are shifted cyclically in semitones and the shifting steps are counted and the chord patterns shifted in this way each time with the undisplaced one Compares input signal pattern. The first alternative has the advantage that one can manage with a small number of shifting steps. The second alternative has the advantage that only a single chord pattern per chord type has to be compared, albeit with a longer computing time, whereas in the first alternative, for example with a major triad, a total of three chord patterns associated with this triad are added to the input signal pattern are comparing.

In order to be able to replace the input signal pattern with a correspondingly corrected input signal pattern when one of the chord patterns is present, the chord patterns of the predetermined amount of chord patterns are assigned signal patterns which can either already correspond to the corrected input signals (for example by specifying the respective tone frequency or, preferably, correction signals for form the input signals.

So that the signal patterns assigned to the chord patterns can also be limited to an octave, it is proposed, for simple consideration of the initial shift of the input signal pattern or the chord patterns, that if the input signal pattern within the definition octave matches a chord pattern of the specified number of chord patterns loads the respective chord pattern associated signal pattern, limited to one octave, into an output memory and, depending on the number of shift steps, cyclically shifts the signal pattern in the output memory in semitones in the opposite direction, if necessary. The direction of shift is therefore opposite to the initial shift of the input signal pattern or the chord pattern. In the case of cyclic shifting of the chord patterns within the octave, ie with the storage of the memory content from the overflow end of the memory to at the other end of the memory, there is accordingly a cyclical shift in the opposite direction in the output memory. The correction signals are then in the corresponding position of the octave, so that now only the input signals, regardless of which of the possible octaves they are, need to be corrected. The correction signals preferably relate to relative frequency changes, in particular given in cents, in order to achieve independence from the octaves.

A particularly preferred development of the method according to the invention is characterized in that, after ascertaining an input signal pattern corresponding to a chord pattern, it is determined in the subsequent input signal patterns whether the corresponding tone or the corresponding tones of the input signal pattern are completely contained in the chord pattern and, if applicable, this tone or these tones corrected according to the chord pattern. This measure has the advantage that immediately after playing a chord from the set of patterns and sounding in a harmonious mood, its individual tones or combinations of single tones of this chord can be played without changing the mood of these tones. This is very advantageous if, for example, a chord is played a chord for intonation and then the individual notes of this chord are to be played.

In a further development of this measure, it is proposed that additional chord tones be assigned to the pattern chords, and that if an input signal pattern corresponding to a chord pattern is determined in the subsequent input signal patterns, it is determined whether the corresponding tone or the corresponding tones of the input signal pattern correspond to additional chord tones of the established chord pattern and, if applicable, this tone or this Tones corrected according to the additional chord tones. These additional chord tones are tones that follow the pattern chord downwards or upwards. Large or small thirds are preferably provided, with a large third of the pattern chord being followed by a small third for the additional chord tone and vice versa. In the case of a major triad, there is an additional chord note at the bottom of a minor third and an additional chord note at the top of a major third.

In the case of a multi-manual input device, it is proposed that the associated input signal pattern be compared separately with the chord patterns of the predetermined amount of chord patterns in each manual. Since accompaniment chords are generally played on one of the manuals, these can be identified even if notes other than chords, for example so-called continuity tones, are played on another manual.

It is further proposed that after determining an input signal pattern corresponding to a chord pattern in one of the manuals, the subsequent input signal patterns in all manuals are checked to determine whether the corresponding tone or tones are completely contained in the chord pattern. Thus, the chord-identical tones of all manuals are corrected, while the tones not belonging to the determined chord pattern maintain the frequencies of the tempered mood.

Instead of or in addition to the proposed additional correction of the fundamental frequencies of the chords, in order to avoid discordant frequency differences of the same tones of successive chords, it is proposed that two successive input signal patterns, each corresponding to two different chord patterns, correspond to the specified number of chord patterns. determines whether the same tone occurs in both chord patterns and, if applicable, makes such an additional correction to the second input signal pattern that the matching tones in both chords have essentially the same level or at least have a frequency difference that does not exceed a predetermined value of preferably less than 8 cents. If this additional correction only affects the matching tone, there is a slight deviation of this tone from the variable tuning in the second chord. However, it is also possible to send this additional correction to all tones of the new chord, so that there is a shift of all tones of this new chord to "higher" or "lower". This would maintain the frequency relationships of the variable tuning. Since such a shift to "higher" or "lower" could occur several times in succession in the same direction in exceptional cases, provision is preferably made to limit this additional correction, preferably to a value of less than 16 cents in one direction overall.

The invention also provides in claim 16 a pitch control for a musical instrument for performing the method described above.

As the signal pattern stored in the chord pattern storage circuit, the desired chord in the variable tuning can be matched directly, e.g. signals indicating their frequency are stored. However, it is particularly preferred that coordinate signal patterns are stored in the chord pattern memory circuit as signal patterns for correcting the note input signals according to the variable mood, and that a correction circuit is provided to which the note input signals and the correction signals of the correction signal patterns can be applied, and which as output signals that the note input signals, which have been corrected in accordance with the correction signals, are sent to the sound generating device. This embodiment of the invention leads to a simplified structure of the control, in particular because it allows the memory requirement of the chord pattern memory circuit to be reduced (12 memory locations per correction signal pattern).

In order to simplify the procedure in the chord recognition circuit with a reduced memory requirement, it is proposed that a Definition octave memory is provided with 12 memory locations which are assigned to the 12 different tones of a given octave, whereby when checking an input signal pattern corresponding to a chord, a memory location is occupied if the tone corresponding to this memory location occurs in the chord in any octave. Due to this projection of the input signal pattern onto the definition octave, only a correspondingly reduced amount of information has to be worked with.

It is also proposed that a working memory be provided with 12 storage locations, into which the memory content of the definition octave memory can be transferred, and a shift counter, which, starting from the counter value "O", is increased by "1" when the memory content of the working memory is increased by a memory location is moved in a predetermined direction.

Accordingly, a chord pattern memory can be provided, each with a memory line assigned to one of the chord patterns of the predetermined number of chord patterns, in particular each with 12 memory locations. Due to the above-mentioned projection of the input signal pattern onto the definition octave, the chord patterns to be compared herewith can also be limited to one octave (12 memory locations). If the main memory is designed as a shift register memory, the respective memory content can be shifted in the predetermined direction until a occupied memory space corresponding to a chord tone has reached the corresponding end of the main memory. The “edge-adjusted” chord pattern from the chord-pattern memory is then compared in turn with the “edge-adjusted” input signal chord in this way.

Furthermore, a chord memory can be provided with 12 memory locations assigned to each tone of an octave for storing the last recognized chord. This makes it possible to refrain from comparing the chord pattern if the same chord is played several times in succession. In addition, this comparison of the tones played with the chord memory has the advantage that the frequency of the tones does not change if, after a recognized and frequency-corrected chord, chords are subsequently played from a subset of the tones of the previous chord or individual tones of this chord.

Furthermore, a correction memory can be provided with memory lines, in particular with 12 memory locations assigned to each of the 12 different semitones of an octave, each of which is assigned to a chord pattern of the predetermined number of chord patterns.

It is also proposed that an output memory be provided, in particular with 12 memory locations assigned to each of the 12 different semitones of an octave, in which the content of the memory line of the correction factor memory assigned to a recognized chord can be transferred, and the memory content of which can preferably be shifted in a predetermined direction is. In this way, the edge adjustment carried out to facilitate the comparison of the chord pattern when the correction factors are output can be taken into account in a simple manner by a corresponding shift back in the output memory.

The invention is explained below with the aid of tables designated I - VI at the end of the description and with the aid of the drawing.

Table I gives the designation of the function of the tones of a number of selected chords in the representation chosen here.

Table II shows the frequency relationships of the tones of a chord to each other, both in the harmonic mood and in the tempered mood.

Table III shows a list of the chords to be corrected with the assigned correction values.

Table III A shows a list according to Table III with modified correction values.

Table IV shows the associated chord patterns stored in the chord recognition memory for each of the selected chords.

Table V assigns the chord pattern numbers according to Table IV to the note examples according to FIG. III.

Table VI shows the effect of a halftone chord shift.

1 shows an example of a note (transition from an E minor chord to a C major chord) together with a table to explain the additional correction.

Fig. 2 shows a greatly simplified circuit diagram.

Fig. 3 shows a series of note examples labeled a - n.

3a shows further note examples α and β.

4 shows the assignment of a number of memories used for the method according to the invention.

5 shows the upper half of a flow chart.

6 shows the lower half of the flowchart mentioned.

entsprechend 100 Cents. Es sind jedoch auch andere feste Ausgangsstimmungen denkbar, wie z.B. die in der DE-PS 25 58 716 angegebenen Stimmungen. Um verschiedene Tonarten spielen zu können und auch Tonart-Modulationen durchführen zu können, ist es bei Instrumenten, die,im Gegensatz beispielsweise zu Streichinstrumenten und Blasinstrumenten/während des Spiels vom Spieler nicht nachgestimmt werden können, erforderlich, eine derartige feste Stimmung vorzusehen, so bei den Tasteninstrumenten Klavier und Orgel (mit Pfeifen oder elektronisch). Gemäß der Erfindung soll nun für Instrumente, welche mehrstimmiges Spiel erlauben und somit das Spielen von Akkorden, selbsttätig eine Frequenzkorrektur der Töne derart vorgenommen werden, daß die Akkorde harmonisch rein gestimmt sind. Das Instrument muß hierzu eine Tonerzeugungseinrichtung aufweisen, welche die Erzeugung von frequenz-korrigierten Tönen zuläßt. Die-se Voraussetzung ist bei "elektronischen Orgeln" bzw. "Synthesizern" von vorneherein gegeben. Es ist jedoch auch denkbar, eine Pfeifen-Orgel einzusetzen, bei welcher jedem einzelnen Ton mehrere Pfeifen unterschiedlicher Tonhöhe (z.B. Länge) zugeordnet sind mit wahlweiser Ansteuerung der jeweils gewünschten Pfeife. Auch ist eine Pfeifen-Orgel einsetzbar, bei welcher Pfeifen mit variabler Tonhöhe (z.B. variabler Länge) verwendet werden, um während des Spiels die gewünschte Nachstimmung der Pfeife zu ermöglichen.In the inventive method for pitch control described in more detail below, a fixed tuning is assumed, which corresponds to the tempered tuning with division of an octave into 12 identical semitones, that is to say with a frequency ratio of $\text{¹²}\sqrt{\text{2nd}}$

corresponding to 100 cents. However, other fixed initial moods are also conceivable, such as the moods specified in DE-PS 25 58 716. In order to be able to play different keys and also to be able to perform key modulations, it is necessary to provide such a fixed tuning for instruments which, in contrast to, for example, string instruments and wind instruments / cannot be re-tuned by the player during the game the keyboard instruments piano and organ (with pipes or electronic). According to the invention, a frequency correction of the tones should now be carried out automatically for instruments which allow polyphonic playing and thus the playing of chords in such a way that the chords are harmonically pure. For this purpose, the instrument must have a tone generating device which permits the generation of frequency-corrected tones. This is a prerequisite for "electronic organs" or "synthesizers" right from the start. However, it is also conceivable to use a pipe organ in which several pipes of different pitch (eg length) are assigned to each individual tone, with optional activation of the desired pipe. A pipe organ can also be used, in which pipes with variable pitch (eg variable length) are used in order to enable the desired tuning of the pipe during the game.

Table V indicates those chords that are suggested for the harmonically pure tuning, depending on the application, less important chords can be omitted or additional chords can be added. Also, chords with more than four different notes are used in the embodiment not considered. All chords are recognized and corrected not only in their basic position according to Table I (root note G as the lowest note), but also in all reversals, positions and doubles. This is achieved in that all tones from all octaves are projected onto an octave designated as a "definition octave", for example consisting of the 12 consecutive semitones from c 'to h'. 3, the tone C is the lowest tone when projected onto the definition octave from c 'to h', the tone E in the C major chord denoted by a1 is the next higher and the tone G the highest on this definition octave, so that this chord appears on the definition octave exactly as shown and can be identified by chord pattern No. 1 according to Table IV. The chord a2 is an A flat major triad, in which its tones, projected onto the above-mentioned definition octave, would be read from bottom to top in the order C, Eb and A flat, which corresponds to the chord pattern No. 2 according to Table IV. Accordingly, the F major triad is read as example a3 from bottom to top in the order C, F and A and corresponds to chord pattern No. 3 from Table IV.

The chords played appear when they are projected onto the definition octave in their basic position or in one of their inversions, regardless of the position, doubling or inversion in which they are played. The position of the chord in the definition octave does not have to correspond to the position in which the chord is played, but depends on the beginning ° and end tone of the selected definition octave and on which concrete tones the chord played consists of. As further shown in Table IV, the representation of the tones on the definition octave and the specific grid of each chord enable the function of the individual tones of the chord (here as letters G, M, T, Q, R and S according to Table I) can be determined and so each tone with a certain function in the chord can be assigned a very specific correction value from tempered to harmonic tuning.

On the basis of FIG. 2 in conjunction with Table II, it is to be demonstrated that audible intolerances can occur even with individually harmonically pure tuning of successive chords. 1 shows the transition from an E minor triad to a C major triad. In order to have a fixed reference system, for example, the respective chord tone is defined in the function G in the fixed (tempered) mood, i.e. with the fundamental tone e (= 330 heart). The tones g (with function M) and h (with function Q), which would accordingly have frequencies of 392 (= 300 cents) and 494 hearts (= 700 cents) if the mood was tempered (see Table II), are then set to Frequencies 396 heart and 495 heart corrected. Accordingly, the root note c (with function G) of the C major chord is set to 523 hearts with harmonic frequencies 392 and 327 for the lower notes e (with function Q) and g (with function T).

If you now compare the two lowest tones of both chords, which both correspond to the same tone, namely the tone e, these tones played in direct succession have a frequency difference of approx. 1% due to the correction to the harmonic mood. This also applies to the two middle tones of the chords with 396 and 392 hearts. Such a frequency difference of tones played directly on top of one another, which are in themselves the same, is audible and is perceived as unpleasant.

To avoid this effect, the harmoniously tuned major triads as a whole are shifted to higher frequencies and, accordingly, those are still harmoniously tuned minor triads to lower frequencies. According to Table III, a frequency shift of the major triads upwards by 4 cents and a frequency shift of the minor triads downwards by 6 cents have been found to be particularly advantageous.

Due to this additional correction, the frequency difference of the respective tones is now less than 1/3% with 1 heart each and is therefore no longer audible.

In Table III, the third column from the right shows the preferred additional corrections for the specified chords.

Table III A shows alternative additional corrections for a number of chords which are characterized by improved fifths purity.

2 shows a purely schematic circuit diagram to explain the method. An input device 10 for inputting note input signals in the fixed mood is symbolized as a series of piano keys 12 for actuating switches 14 assigned to each key 12. The lines 16 emanating from the switches 14 are combined to form a collecting line 18. A sound generating device 20 has a sound signal output circuit 22, which is generally provided with sound frequency generators and which drives one or more loudspeakers 26 via a line 24. Instead of the loudspeaker 26, a recording device, such as a tape, can also be provided for "music buffering". The line 18 opens into a chord pattern recognition circuit 28, from which in turn a line 30 extends for connecting the circuits 28 and 22. The input device 10 and the tone generating device 22 correspond in structure and function to the corresponding components of conventional electronic keyboard instruments.

The chord pattern recognition circuit 28 is connected via a line 31 to a control circuit 32, which in turn is connected via a line 33 to a signal pattern storage circuit 34. The control circuit 32 is additionally connected via a line 35 to the audio signal output circuit 22.

The general function of the circuit arrangement according to FIG. 2 is the following:
The note input signals are fed via line 18 to chord recognition circuit 28. The chord recognition circuit checks whether an input signal pattern corresponding to a chord from several different tones corresponds to a chord pattern from a predetermined number of chord patterns. If this is the case, this is reported to the control circuit 32, which retrieves the correction signals assigned to this chord from the signal pattern memory circuit and forwards them via line 35 to the audio signal output circuit 22, which accordingly correspondingly transmits the note input signals to it via line 30 corrected and as corrected output signals to the speaker 26.

The method described is of course not limited to such an electrical circuit arrangement, but can also be implemented by appropriately programmed program-controlled devices.

A corresponding program sequence is again shown purely schematically in FIGS. 4 and 5. The memories addressed in the program flow diagram are explained in more detail in FIG. 4. One can see a definition octave memory 40 with twelve memory locations 42, each of which has a semitone of the scale, for example starting with tone c. Has a chord memory 44 same structure. A working memory 46 also has twelve storage locations; however, the working memory 46 is designed as a shift register memory, so that the memory locations are numbered from 1 to 12 and no tone of the scale is assigned. A shift counter 48 is assigned to the working memory 46 and counts the shifting steps carried out in each case by one storage space, corresponding to a semitone of the scale.

A chord recognition memory 50 each has a memory line 52 assigned to the chord pattern according to Table IV, each with twelve memory locations 54. As a comparison, for example of the first four memory lines, with Table IV reveals chord patterns No. 1-4, the memory location allocation in chord recognition corresponds Memory 50 the chord patterns. Thirty-nine lines 52 are provided for the chords proposed for correction here.

A correction factor memory 56 is also organized in thirty-nine lines 58, each with twelve memory locations 60. While either a "1" (ie chord tone) or a "0" (ie no chord tone) appears in the chord recognition memory corresponding to the respective chord pattern, in the correction factor memory 56 there are those memory locations which have the "in chord recognition memory 50 with" 1 "provided corresponding storage locations correspond to the correction signals assigned to the respective tone in accordance with Table III. These correction signals correspond to the total correction in cents from the second column from the right of Table III. For example, if you look at the third line in the correction factor memory 56, which is assigned to the chord pattern No. 3, the number "6" is stored in the first memory location - this is because this tone according to Table IV functions according to the tone with the function Q (= fifth) corresponds to which note again according to Table III, top row a correction of +6 cents is assigned. The memory locations 50 corresponding to the locations 50 of the memory 56 occupied by "0" are also assigned a "0".

Finally, an output memory 62 is also provided, again with twelve memory locations, which are numbered to indicate that this memory is also designed as a shift register memory.

2, the memories 40, 44, 46 and 50 can be assigned to the circuit 28, the memory 56 to the circuit 34 and the memory 62 to the circuit 32.

The process sequence or program sequence can be seen from FIGS. 5 and 6. Starting from the start block 70, the next decision block 72 checks whether the input signals emitted by the tone generator 10 are unchanged, i.e. whether the current switching status remains, for example one or more keys are pressed unchanged. If this is the case, the program jumps to block 74, which will be explained later, and then to "return block" 76. The result is the unchanged output of the input signals corrected as before to the tone generating device 20, so that the tones just played are unchanged Mood continues to ring.

If, on the other hand, the input signals are changed, the input signal pattern is loaded into the definition octave memory 40 according to a block 84, in such a way that, for example, the memory cell 42 assigned to the tone c is assigned "1" if one or several keys assigned to the tone c in any octave are pressed. Otherwise, the memory locations get the Memory content "0". As a result, tones of the same name of any octave are linked by the logical function "or", so that the desired projection of the entered chord on the definition octave is obtained.

In the following block 86 it is checked whether the chord now struck consists exclusively of chord tones of the chord struck last and recognized as a chord pattern. If this check in decision block 88 reveals that there is a pure repetition, the procedure is shortened to block 74, with the result that the new chord sounds with the frequency corrections corresponding to the last played chord, the new "chord" also being off can only consist of a single chord tone of the previously played chord recognized as a chord pattern.

However, if the new chord deviates from this previously played chord in at least one tone, the program proceeds to a decision block 89, in which it is checked whether the input signal pattern corresponds to only a single tone. If this is the case, the program proceeds to a block 80, in which the output memory 62 already mentioned is deleted, as does the chord memory 44 in a subsequent block 82, whereupon the program again proceeds to a block 74 for output of the single tone corresponding input signals to the sound generating device 20 without a correction, since the correction factors in the output memory are set to "0". The single tone therefore sounds in a tempered mood.

If, on the other hand, it is determined that the input signal pattern corresponds to several tones, the content of the definition octave memory 40 is loaded into the working memory 46 in accordance with a block 90. The shift counter 48 is set to the number "0" in a subsequent block 92. The The next program loop serves to shift the memory content of the main memory until a "1" has reached the left edge of the shift register-like memory line forming the main memory 46, for example. This can also be called edge adjustment. In this way, the comparison with the chord patterns in the chord recognition memory is to be facilitated, since the content of the corresponding lines 52 of this memory 50 is also edge-adjusted, as can be seen in FIG. 4.

In a decision block 94 following block 92, a check is carried out to determine whether there is a "1" in memory cell No. 1 of main memory 46. If this is not the case, the program proceeds to block 96 in order to shift the content of the working memory 46 to the left by one cell (corresponding to a semitone). At the same time, the storage value of the shift counter 48 is increased by "one" in block 98. The program then returns to decision block 94. The loop formed in this way is traversed until the edge adjustment is reached, i.e. a "1" is stored in the memory cell 1.

The program then continues to block 100 (FIG. 6). In this case, the chord recognition memory 50 is actuated, namely its first line with the chord pattern No. 1. In the subsequent program loop, the edge-adjusted content of the working memory 46 is compared in turn with all the chord patterns until either equality with a specific chord pattern has been determined or until all chord patterns have been mismatched. In a block 102 of the loop, the respective chord recognition pattern is compared with the content of the working memory. In a subsequent decision block 104, a transition is made to a next decision block 106 within the loop, if the current chord recognition pattern does not match the content of the working memory. In decision block 106 it is checked whether all chord patterns have already been checked. If this is not yet the case, ie the current chord pattern number is smaller than the highest chord pattern number (in the example according to FIG. 4: 39), the program proceeds to a block 108, in which the next one is caused Line of the chord recognition memory 50 is driven. The program then returns to block 102 within this loop.

On the other hand, if it is determined in decision block 104 that the chord being played corresponds to a pattern chord, i.e. the content of the working memory corresponds fully to that of a memory line of the chord recognition memory, the program leaves said loop and proceeds from decision block 104 to a block 110, according to which the content of the chord memory 44 is updated by taking over the content of the definition octave memory 40.

A block 112 follows, according to which the memory line of the correction factor memory 56 is activated, the number of which corresponds to the currently activated line of the chord recognition memory 50, that is to say the number of the chord pattern which has been found to be identical to the currently played chord. This line is copied to the output memory 62 in a subsequent block 114.

Corresponding to the chord recognition memory 50, the contents of the memory lines 58 are also edge-adjusted in the correction factor memory 56. In order to be able to assign the edge-adjusted correction factors determined in this way during tone generation to the played chord tones in their non-shifted position, the edge adjustment of these correction factors in the output memory 62 is reversed. For this serves a program loop following block 114. Subsequent to block 114, as part of the loop, a decision block 116 is approached, in which it is checked whether an edge adjustment in the loop formed by blocks 94, 96 and 98 had to be carried out at all. If the chord in the definition octave memory has been edge-adjusted from the outset (ie in the case of a played chord containing the tone c, provided the definition octave begins with the tone c), a shift in the working memory is of course not necessary. In the latter case, the shift counter would still have the value "0". If this is not the case, block 116 is followed in the loop by block 118, according to which the memory content of the output memory 62 is shifted as a whole to the right, that is to say to the next higher cell number. The value in the shift counter is then decreased by one in a block 120. The program loop then returns to decision block 116. The loop is therefore repeated until the shift counter has the value "0", so that the correction factors in the output memory are in the same position as the tones of the definition octave.

For the memory allocation of the output memory 62, it should be added that if, for example, it is determined that the chord pattern No. 4 is present, the output memory accordingly at positions 2, 3, 5, 6, 7, 9, 10, 11 and 12 each F = Has O and at position 1 F = -6, at position 4 F = 10 and position 8 F = -4. F = O means that there is no pitch correction to be made for the tone in question, that is to say it sounds in a tempered mood. Otherwise the tone is corrected according to the specified correction factor (in cents).

Since the initial shift is within the octave of memory (in the loop with blocks 94.96 and 98) later reversed in the loop 116, 118, 120 within the octave of the output memory (with an identical chord pattern), there is no risk of the output memory overflowing, that is to say that a non-zero correction factor is pushed out of the memory .

However, it is also conceivable to carry out the chord recognition in such a way that a single chord pattern, for example the chord pattern No. 1 for the major triad, is used for each chord, which is then cyclically shifted in a sliding memory with twelve storage spaces, so that with it chord patterns 2 and 3 are also available in the meantime (chord pattern 2 results, for example, from a cyclical shift of chord pattern No. 1 in Table IV to the left by four semitones). It is then necessary to shift the one chord pattern for each chord by a full cycle (12 steps) and to compare each time with the chord played, it being not necessary to adjust the edge of this chord. With this procedure, the output memory would then have to be organized cyclically with a shift in the opposite direction, corresponding to the number of shift steps required to match the chords.

It should be briefly referred to Table VI, which shows that (in the case of a definition octave beginning with the tone c) a played E major triad requires four shifting steps in the working memory 46 to the left until the edge adjustment is reached. Accordingly, the memory content of the output memory 62 corresponding to line No. 4 of the correction factor memory 56 must then be shifted to the right by four steps, so that the correction factor "-6 cents" is then at the memory location assigned to the tone e.

After carrying out the necessary shifts in the output memory (value in the shift counter = zero), said loop is exited; the program goes from decision block 116 to block 74. The input signals are now corrected in accordance with the correction factors in the output memory. Regardless of the octave in which the respective tone is located, the correction factor corresponding to its name in the output memory for this tone is used to correct this tone. A kind of back-projection onto the original multi-octave input signal pattern is thus carried out. Since the correction factors indicate frequency ratios, they are octave-independent. In many common audio signal output circuits 22, an audio frequency generator is assigned to each key 12 from the outset. According to the invention, controllable tone frequency generators are to be provided which, based on the tempered basic mood, can be re-tuned automatically on the basis of the correction factors.

As a result, a harmonically corrected chord is output by the tone generator 20 when it is determined that this chord corresponds to a predetermined chord pattern. If the chord cannot be recognized, the chord is generated in a tempered mood. For this purpose, a second loop output is provided in the program loop comprising blocks 102, 104, 106, 108, namely in decision block 106. If it is determined in block 106 that on the one hand the chord played does not match the current chord pattern (block 104) and on the other hand that already highest chord pattern number (eg 39) is reached, the program proceeds from block 106 to a block 122, according to which all correction factors for the output memory 62 are set to zero cents.

In a subsequent block 126, the chord memory 44 is deleted. The program then goes back to block 74, that is to say to the output of the uncorrected input signals in this case to the sound generating device 20. The program then returns to the start of the program (block 70) via the "return block" 76. The entire program loop can be run through independently of a key actuation of the instrument with a fixed repetition frequency.

According to the above, chord patterns are automatically identified and their individual tones are corrected immediately, so that the chord that is sounded harmoniously pure. Single notes or chords struck later that are part of the last identified chord pattern are also corrected. However, you can also go further and assign additional chord tones to the individual pattern chords that are tuned in relation to the actual chord tones. If one of the additional chord tones is struck after identifying this chord, it is also corrected accordingly.

FIG. 3A bears, for example, a pattern chord labeled α (C major triad), which is expanded upwards by an additional chord tone at grade h at a distance of a major third from the top pattern chord tone and downwards by an additional chord tone (tone a ) at a distance of a minor third. These additional chord tones are symbolized in the chord β in FIG. 3A as marked notes. The result is an alternating sequence of major and minor thirds.

The correction factors associated with these individual tones in comparison to the tempered mood are given in FIG. 3A to the right of the chord β in cents.

undsoweiter.If the chord pattern α is identified in the course of a game, both its tones are corrected immediately, so that this chord sounds harmoniously pure; moreover, if one or more of the tones of the chord β which also contains the additional chord tones are subsequently struck, they are each harmonically corrected.

and so on.

Claims (23)

1. Method for automatic pitch correction according to a harmony-dependent variable tuning, especially the harmonic tuning, for a musical instrument with an input device (10) for the input of note input signals in a predetermined fixed tuning, especially the tempered tuning, and with a tone-generation device (20) to which the note input signals are applicable, with the following steps:

   a) in the case of an input signal pattern corresponding to a chord, one determines by comparison with pattern chords of a pre-determined quantity of chord patterns whether one of these chord patterns is present;

   b) if a chord pattern is present the input signal pattern is replaced by an input signal pattern corrected according to this chord pattern and applied to the tone generation device (20), whereby the signal of the input signal patterns allocated to a predetermined fundamental note of the respective chord pattern is fixed according to the predetermined fixed tuning, and the signals, allocated to the other notes of the chord patterns, starting from the fundamental, are corrected in acoordance with the variable tuning,

   characterized in that, additionally, the signal of the input signal patterns allocated to the fundamental note is corrected in relation to the predetermined, fixed tuning and the signals of the input signal pattern, allocated to the other notes of the chord pattern, starting from the corrected fundamental note, are corrected in accordance with the variable tuning, in which method the signal allocated to the fundamental note of a chord is corrected in such a way that the correspondingly corrected note delivered by the tone generation device lies higher or lower than the fundamental note in the predetermined fixed tuning, according to whether the chord notes corrected according to the input ignal pattern lie on average lower or higher than the uncorrected chord notes in the fixed tuning.
2. Method according to claim 1,characterized in that the signal allocated to the fundamental note of a chord is corrected in such a way that the displacement of a mean frequency of the chord notes by reason of the correction of the chord notes is at least approximately compensated by the input signal pattern.
3. Method according to claim 2, characterized in that the signal allocated to the fundamental note of a chord is corrected with an additional correction signal giving a relative frequency variation, which correction signal substantially corresponds to the mean value of the correction signals, which likewise indicate relative frequency variations, for the input signals, in amount, but with reversed sign.
4. Method according to claim 3, characterized in that the input signal pattern is projected onto a predetermined octave (definition octave) and compared with the chord patterns, likewise limited in each case to one octave, of the pre-determined quantity of chord patterns.
5. Method according to claim 4, characterized in that the input signal pattern is shifted within the definition octave, as a whole by half-notes, and the shift steps are counted, until a signal is present at a predetermined end of the definition octave, and in that the input signal pattern shifted in this way is compared with the chord patterns of the pre-determined quantity of chord patterns, and in the case of the chord patterns likewise a chord note is present at the pre-determined end of the octave.
6. Method according to claim 4, characterized in that the chord pattern of the pre-determined quantity of chord patterns is shifted cyclically with the octave by half notes, and the shift steps are counted, and the chord patterns shifted in this way are compared with the unshifted input signal patterns.
7. Method according to either claim 5 or 6, characterized in that in the case of conformity of the input signal pattern within the definition octave with the chord pattern of the pre-determined quantity of chord patterns a signal pattern, allocated to the respective chord pattern and limited to one octave charges into an output store (62), and shifts the signal pattern by half notes in the output store (62) in the opposite direction, possibly cyclically, in accordance with the number of the shift steps of the input signal pattern or of the chord pattern.
8. Method according to one of the preceding claims, characterized in that correction signals for the input signals are used as signals of the input signal patterns.
9. Method according to claim 8, characterized in that the correction signals indicate relative frequency variations.
10. Method according to one of the preceding claims characterized in that after ascertaining an input signal pattern corresponding to a chord pattern, one ascertains among the following input signal patterns whether the corresponding note or notes of the input signal pattern is or are completely contained in the chord pattern and, if so, corrects this note or these notes according to the chord pattern.
11. Method according to claim 10, characterized in that additional chord notes are allocated to the pattern chirds, and in that on ascertaining an input signal pattern corresponding to a chord pattern among the following input signal patterns one ascertains whether the corresponding note or notes of the input signal pattern correspond to additional chord notes of the ascertained chord pattern and, if so, corrects this note or these notes in accordance with the additional chord notes.
12. Method according to one of the foregoing claims, characterized in that in the case of a multi-manual input device, the respectively allocated input signal pattern. is compared separately for each manual with the chord patterns of the pre-determined quantity of chord patterns.
13. Method according to one of the preceding claims, characterized in that one compares only input signal patterns which correspond to chords having at least three different notes.
14. Method according to either claim 12 or 13, characterized in that after ascertaining an input signal pattern corresponding to a chord pattern in one of the manuals, the following input signal patterns of all manuals are examined to ascertain whether the corresponding note or notes is or are contained completely in the chord pattern.
15. Method according to one of the foregoing claims or to the preamble to claim 1, characterized in that in the case of two successive input signal patterns which correspond in each case to two different chord patterns of the predetermined quantity of chord patterns one ascertains whether the same note occurs in both chord patterns and if so effects such an additional correction in the second input signal pattern that the coinciding notes of the two chords possess in both chords substantially the same level, or at least possess a.frequency difference not exceeding a predetermined value preferably of less than 8 cents.
16. Pitch control system for a musical instrument for carrying out the method according to one of the foregoing claims with an input device (10) for the input of note input signals in a pre-determined fixed tuning, especially the evenly beating tempered tuning, and with a tone generation device (20) to which the note input signals are applicable, including

    - a chord recognition circuit (28) which at every input signal pattern corresponding to a chord ascertains whether this input signal pattern corresponds to a chord pattern from a pre-determined quantity of chord patterns,

    - a signal pattern store circuit (34) in which a signal pattern is stored for each chord pattern of the predetermined quantity of chord patterns, including a signal allocated to a predetermined fundamental note of the respective chord pattern,

    - a control circuit (32) which, when the chord recognition circuit (28) ascertains that an input signal pattern corresponding to one of the predetermined chord patterns is applied, causes the signal pattern storage circuit (34) to give off the signal pattern corresponding to the ascertained chord pattern to the tone generation device (20) to generate the chord concerned in the variable tuning,

    characterized in that the control circuit includes means which correct both the signal of the input signal pattern allocated to the fundamental note of a chord as compared with the predetermined fixed tuning, and also the signals of the input signal pattern allocated to the other notes of the chord patterns starting from the corrected fundamental note, in accordance with the variable tuning, in such a manner that the correspondingly corrected signal allocated to the fundamental note of a chord leads to the emission of a tone by the tone generation device which lies higher or lower than the fundamental note in the pre-determined fixed tuning, according to whether the chord notes corrected according to the input signal pattern lie on an average lower or higher than the uncorrected chord notes in the fixed tuning.
17. Pitch control system according to claim 16, characterized in that correction signal patterns are stored as signal patterns in the signal pattern storage circuit (34), for the correction of the note input signals according to the variable tuning, and in that a correction circuit (tone signal output circuit (22)) is provided to which the note input signals and the correction signals of the correction signal patterns are applicable, and which delivers as output signals to the tone generation device (20) the note input signals corrected according to the correction signals.
18. Pitch control system according to either claim 16 or 17, characterized in that a definition octave store (40) is provided having 12 storage places (42) allocated to each note of a pre-determined octave, while on examination of an input signal pattern corresponding to the chord a storage place is occupied when the note corresponding to this storage place occurs in the chord in any desired octave.
19. Pitch control system according to one of the preceding claims, characterized in that a work store (46) is provided having 12 storage places into which the stored content of the definition octave store (40) can be transferred, and in that a shift counter (48) is provided which, starting from the counter value "O", is increased by "one" every time when the stored content of the work store (40) is shifted by one store place in a pre-determined direction.
20. Pitch control system according to one of the preceding claims, characterized in that a chord pattern store (50) is provided in each case having a storage line (52) allocated to one of the chord patterns of the predetermined quantity of chord patterns, especially having in each case 12 storage places (54).
21. Pitch control system according to one of the preceding claims, characterized in that a chord store (44) is provided having 12 storage places, allocated to each note of an octave, for the storage of the last-recognized chord.
22. Pitch control system according to one of the preceding claims, characterized in that a correction factor store (56) is provided, especially having storage lines with 12 storage places allocated in each case to each note of an octave, a chord pattern of the pre-determined quantity of chord patterns being allocated to each of the storage lines.
23. Pitch control system according to claim 22, characterized in that an output store (62) is provided, especially having 12 storage places into which the content of the storage line (58), allocated to a recognized chord, of the correction factor store (50) can be transferred, and the stored content of which is displaceable, preferably in a pre-determined direction.

Images